The pathogenic mechanisms underlying the neuronal degeneration and death in ALS are as yet unknown. There are currently several hypotheses concerning the pathogenic processes involved in MND, none of which are mutually exclusive. The main hypotheses include protein misfolding and aggregation, oxidative stress, disruption of axonal transport, and glutamatergic excitotoxicity; there is also evidence that inflammation, autoimmunity and apoptotic cell death pathways are involved.
Cell types involved in MND
Activated/reactive astrocytes and microglia are often found in MND post-mortem tissue (Ekblom et al., 1994; (Kawamata et al., 1992; Schiffer et al., 1996) and in mutant SOD1 transgenic mice (Bruijn et al., 1997b; Cha et al., 1998; Tu et al., 1996), and although the clinical and molecular pathology of MND indicates an obvious involvement of motor neurons, it is increasingly being recognised that glial cells also play an important role in the pathogenesis of the disease. Important insights into the interdependence of different cell types in MND pathogenesis have come from several studies in mutant SOD1 transgenic mice. The discovery that cell-specific expression of mutant SOD1 in either neurons or astrocytes did not cause disease in mice (Gong et al., 2000; Lino et al., 2002; Pramatarova et al., 2001) indicated that both neuronal and non-neuronal cells could play a role in disease pathogenesis. A chimeric SOD1 mouse model was subsequently created, composed of a mixture of normal cells and mutant SOD1-expressing cells in order to investigate the relationship between different cell types (Clement et al., 2003). The presence of wild-type cells delayed disease onset and extended the lifespan of the chimeric mice compared with those overexpressing mutant SOD1 ubiquitously, and more specifically, degeneration and death of mutant SOD1-expressing motor neurons was reduced when the neurons were surrounded with a sufficient number of normal non-neuronal cells, whereas normal motor neurons surrounded by mutant SOD1-expressing non-neuronal cells showed signs of degeneration such as ubiquitinated inclusions. These results indicate that the expression of mutant SOD1 in both neuronal and non-neuronal cell populations, and the interactions between them, is of fundamental importance in MND.
Toxicity of intracellular aggregates
Mutant SOD1 aggregates in familial ALS (FALS)
The mechanism(s) by which mutant SOD1 causes FALS are not known, although a popular hypothesis is that the mutant protein forms abnormal toxic aggregates within the cell, leading to neuronal degeneration and death. It has been suggested that the toxicity of these aggregates may be effected by disruption of axonal transport (Borchelt et al., 1998; Williamson and Cleveland, 1999), sequestration of heat shock proteins and other chaperones (Bruening et al., 1999; Okado-Matsumoto and Fridovich, 2002; Shinder et al., 2001), dysfunction of the proteasome (Hoffman et al., 1996; Urushitani et al., 2002) and/or damage to mitochondria (Jaarsma et al., 2001; Kong and Xu, 1998; Takeuchi et al., 2002).
SOD1 aggregates are found in motor neurons and surrounding astrocytes in FALS post-mortem tissue (Kato et al., 1997; Shibata et al., 1996b); they are also an early indicator of disease, occurring before the onset of symptoms, in mutant SOD1 transgenic mice (Bruijn et al., 1997b; Johnston et al., 2000; Stieber et al., 2000). They are not, however, a characteristic feature of sporadic ALS (Shibata et al., 1996a). Further support for a role of abnormal SOD1 aggregation in FALS pathogenesis comes from the finding that delaying the formation of such abnormal aggregates, which are high molecular weight ‘insoluble protein complexes’ (IPCs), delays the disease onset in mutant SOD1 transgenic mice. This delay in aggregate formation was achieved by the creation of double transgenic mice carrying both human SOD1G93A and chat-GluR2 (in which the GluR2 subunit of the alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor is overexpressed in a cholinergic neuron-specific manner, resulting in a large reduction in calcium-permeability of AMPA receptors) (Tateno et al., 2004).
The way in which mutant SOD1 causes abnormal protein aggregation is unknown, although structural evidence suggests that it may be by means of destabilisation of normal dimers (Hough et al., 2004), and this has subsequently led to the proposal of dimer stabilisation as a potential therapeutic intervention (Ray and Lansbury, 2004). Abnormal mutant SOD1 aggregates are not only present in the cytosol but have also been found in spinal cord mitochondria in both human cases and mouse models of mutant SOD1-mediated FALS (Liu et al., 2004; Pasinelli et al., 2004; Vijayvergiya et al., 2005). Together with the existence of extensive mitochondrial degeneration and vacuolation in mutant SOD1 mice at pre-clinical stages of the disease (Higgins et al., 2002; Jaarsma et al., 2000; Wong et al., 1995), the localisation of such aggregates suggests an involvement of mitochondrial dysfunction in the pathogenesis of MND.
Intermediate filament aggregates in MND
Neurofilament subunits (NF-L, NF-M, NF-H) and peripherin are major components of the neuronal cytoskeleton and are frequently found in MND inclusion bodies (within neuronal cell bodies and axons) (Carpenter, 1968; Corbo and Hays, 1992; Hirano et al., 1984; Migheli et al., 1993; Rouleau et al., 1996). Transgenic mice overexpressing IF proteins such as NF-L (Xu et al., 1993), human NF-H (hNF-H) (Cote et al., 1993), peripherin (Beaulieu et al., 1999) or alpha-internexin (Ching et al., 1999) all develop a motor neuron pathology with IF inclusions. In humans, mutations in NF-H have been reported in SALS patients (Al-Chalabi et al., 1999; Figlewicz et al., 1994; Tomkins et al., 1998), mutations in the NF-L gene have been found to cause the motor/sensory neuropathy Charcot Marie Tooth disease type 2 (CMT2) (De Jonghe et al., 2001; Fabrizi et al., 2004; Georgiou et al., 2002; Jordanova et al., 2003; Mersiyanova et al., 2000; Yoshihara et al., 2002; Zuchner et al., 2004) and two different mutations in the peripherin gene have recently been reported to cause MND (Gros-Louis et al., 2004; Leung et al., 2004).
It has therefore been proposed that the abnormal IF accumulations found in ALS may contribute to the pathogenesis of the disease. It is unknown, however, how these accumulations form and whether they play a detrimental or protective role in ALS. For example, crossing SOD1G37R mice with mice overexpressing hNF-H resulted in an extended lifespan compared with mice overexpressing SOD1G37R alone, and this correlated with the degree of perikaryal NF accumulation (Couillard-Després et al., 1998). Similar results were found when the phenotype of peripherin overexpressing mice was rescued by overexpressing hNF-H (Beaulieu and Julien, 2003). The overexpression of hNF-H shifted the intracellular localisation of NF-H from the axonal to the perikaryal compartment, which suggests that the protective effect may be due to a reduction of axonal NF accumulation or an increase in perikaryal accumulation.
There is evidence that an increase in perikaryal accumulation is more important in protection from degeneration than a decrease in axonal accumulation, as decreasing the axonal NF content and axonal calibre in mutant SOD1 mice (by disruption in one allele of the NF-L gene) did not affect the disease severity or lifespan of the mice (Nguyen et al., 2000). It has been suggested that perikaryal NF inclusions may act as a ‘phosphorylation sink’ in which NFs are preferentially phosphorylated, thereby protecting other proteins that could be detrimental to the cell when hyperphosphorylated by kinases such as Cdk5 (Nguyen et al., 2001). In support of this hypothesis, it has been found that the perikaryal NF accumulations in the SOD1G37R/hNF-H mice are hyperphosphorylated and co-localise with Cdk5; and hyperphosphorylation of tau at Cdk5 sites is reduced in these mice, suggesting that the NFs are preferentially phosphorylated by Cdk5 (Nguyen et al., 2001). In contrast axonal accumulations of IF proteins are thought to have a detrimental effect on the cell. This may be caused by sequestration of essential proteins and organelles such as mitochondria, and/or by disruption of axonal transport.
Alternatively, sequestration of neuronal nitric oxide synthase (nNOS) in NF aggregates has been proposed to occur in neurons of mice overexpressing NF-L, leading to enhanced N-methyl-D-aspartate (NMDA)-mediated calcium influx that may cause neuronal cell death (Sanelli et al., 2004).
It is believed that NF subunit stoichiometry is a major factor in the formation of NF aggregates. This is supported by the finding that the motor neuron pathology found in hNF-H overexpressing mice was rescued by co-expressing human NF-L (hNF-L) at levels that restored the correct NF-L:NF-H stoichiometry. Additionally, reduced NF-L mRNA levels and selective alterations of NF expression have been observed in inclusion-bearing spinal cord motor neurons of ALS patients (Menzies et al., 2002; Wong et al., 2000). Interestingly, in peripherin overexpressing mice, the disease was dramatically accelerated by a deficiency in NF-L (a phenomenon seen in ALS and in normal ageing) (Bergeron et al., 1994; Krekoski et al., 1996).
In ALS patients, biochemical changes indicative of oxidative damage, such as lipid peroxidation, free carbonyls, protein nitration and protein glycosylation, in motor neurons and spinal cord astrocytes suggests the involvement of oxidative stress in the disease (Abe et al., 1995; Beal et al., 1997; Bowling et al., 1993; Niebroj-Dobosz et al., 2004; Shaw et al., 1995b; Shibata et al., 2001). For reviews see (Robberecht, 2000). Furthermore, the elevation of hydroxyl radicals in the spinal cord of pre-symptomatic SOD1 mutant mice has been reported in several studies (Andrus et al., 1998; Bogdanov et al., 2000; Ferrante et al., 1997; Hall et al., 1998; Liu et al., 1998), which suggests vulnerability of a particular subset of motor neurons at pre-clinical stages of the disease. The role of oxidative stress mechanisms leading to cell death in MND has thus been investigated, particularly in relation to mutant SOD1-mediated FALS.
The toxicity of mutant SOD1 is thought to occur by means of a gain of (unknown) function rather than loss of its superoxide dismutase activity. This is suggested by the dominant inheritance pattern of FALS SOD1 mutations, the absence of any mutations causing severe truncations/loss of the protein and the finding that many of the mutants, for example D90A, retain their dismutase activity despite causing ALS. Although protein aggregation is a widely-favoured hypothesis for toxicity caused by mutant SOD1 in FALS, oxidative damage to the cell by mutant SOD1 has also been suggested as a pathogenic mechanism in FALS. It has also been postulated that the presence of such protein aggregates could be a consequence of reactive oxygen species (ROS) generation and oxidative modifications of the proteins (Gélinas et al., 2000; Rakhit et al., 2002; Valentine and Hart, 2003).
Studies using the spin trapping molecule 5’,5’-dimethylpyrroline-N-oxide (DMPO) (which reduces oxidative stress propagation in NSC-34 motor neuron-like cells overexpressing SOD1G93A) (Liu et al., 2002) revealed that mutant SOD1 can become a source of free radicals by means of enhanced peroxidase activity, and can induce oxidative damage leading to lipid peroxidation, mitochondrial dysfunction and cell death (Liu et al., 2002; Wiedau-Pazos et al., 1996; Yim et al., 1996). This peroxidase activity has not, however, been observed in all studies (Singh et al., 1998). Alternatively it has been suggested that mutant SOD1 causes oxidative stress by production of nitronium ions, which may then nitrate tyrosine residues (Beckman et al., 1994), and this may be due to the mutants’ decreased affinity for zinc (Crow et al., 1997). This is supported by the presence of free 3-nitotyrosine in both human FALS cases and in mutant SOD1 mice (Beal et al., 1997; Ferrante et al., 1997). These nitrotyrosines appear not to be protein bound, however, which calls into question the significance of this finding (Bruijn et al., 1997a), although it has recently been found that free nitrotyrosine can induce apoptosis (Peluffo et al., 2004). However, both of the above oxidative damage hypotheses (enhanced peroxidase and nitration) require the presence of an active-site copper bound to SOD1, but MND still occurred in a transgenic mouse expressing a mutant in which all four copper-coordinating histidines were mutated (Wang et al., 2003). Furthermore, the disease phenotype seen in several mutant SOD1 mouse models was unaffected when the copper chaperone for SOD1 (CCS; a protein that is essential for copper loading of SOD1) was eliminated (Subramaniam et al., 2002).
Although the origins of oxidative stress in MND are not clear, the localisation of mutant SOD1 in the intermembrane space (IMS) (Liu et al., 2004) and matrix (Vijayvergiya et al., 2005) of mitochondria could cause mitochondrial dysfunction (membrane depolarisation, decreased activity of respiratory complexes and Cytochrome c release), and this may contribute to oxidative stress pathways leading to neuronal death in FALS (Beretta et al., 2003; Higgins et al., 2002; Jaarsma et al., 2001; Sturtz et al., 2001). Mitochondrial dysfunction is also implicated in Hereditary spastic paraplegia, as two of the known mutations that cause the disease (SPG7/paraplegin and SPG13/HSP60) are found in genes encoding mitochondrial chaperone proteins, which are involved in protein folding in the IMS. Indeed, several patients with the SPG7/paraplegin mutation show signs of impaired mitochondrial oxidative phosphorylation, which is shown by the presence of abnormal mitochondria (ragged red fibers) and Cytochrome c oxidase deficient-fibers in muscle biopsies (Casari et al., 1998). Furthermore, mitochondrial abnormalities such as hypertrophy, concentric cristae, herniations and giant mitochondria, found in the synaptic terminals of spinal cord motor neurons, is the earliest pathological feature seen in Paraplegin knockout mice (Gelbard, 2004). Together these studies suggest an involvement of mitochondria in the pathogenesis of the MND.
Defects in axonal transport
The presence of IF accumulations in ALS has led to the proposal that NF transport is somehow perturbed, and subsequent studies in mouse models of MND have indicated that defects in slow axonal transport is one of the earliest pathological events in the disease (Williamson and Cleveland, 1999). In particular, axonal transport is abnormally reduced in mice with the SOD1 G93A, G37R and G85R mutations, and this precedes the onset of neuropathology (Borchelt et al., 1998; Warita et al., 1999; Williamson and Cleveland, 1999; Zhang et al., 1997). Additionally, axonal transport defects have been observed in the Wobbler mouse (Mitsumoto et al., 1990; Mitsumoto and Gambetti, 1986), pmn mutant mouse (Sagot et al., 1998), hNF-H overexpressing mice (Collard et al., 1995) and the Paraplegin knockout mouse (Ferreirinha et al., 2004).
It has been hypothesised that NF accumulations in transport-deficient axons may impede further transport along microtubules, leading to cell death by ‘axonal strangulation’ (Williamson and Cleveland, 1999). Axonal transport defects have also been proposed to contribute to the dying back axonopathy observed in HSP due to impairment of efficient protein delivery to synaptic terminals. Furthermore, a lack of trophic support has been implicated in many forms of MND and this may also be caused by faulty transport mechanisms. Motor neurons are one of the largest cells in the nervous system (cell body diameter of 50-60 µm) with axonal processes of up to 1 metre in length and possess a high NF content due to their need for a robust cytoskeletal network. Therefore, defects in axonal transport and abnormal NF accumulation could explain the selectivity of motor neuron death in MND. It is not known how the slowing of axonal transport in MND occurs, although several mechanisms have been proposed.
Role of hyperphosphorylation of NF subunits
Hyperphosphorylation of NF subunits has been proposed to play a role in deregulation of NF transport in MND. The mutations in NF-H that have been identified as a risk factor for SALS affect phosphorylation of side-arm domain Lys-Ser-Pro (KSP) repeat motifs (Al-Chalabi et al., 1999; Tomkins et al., 1998) and such KSP repeat phosphorylation is believed to be important in regulating NF transport. KSP repeats in NF-M and NF-H are phosphorylated by multiple kinases such as glycogen synthase kinase-3α and 3β (Gsk3α/β), p42/p44 mitogen-activated protein kinases (p42/p44) MAPKs, also known as extracellular signal-related kinase (ERK2/ERK1), stress activated protein kinases (SAPKs) and the cyclin-dependent kinase-5/p35 complex (Cdk5/p35) (Brownlees et al., 2000; Giasson and Mushynski, 1998; Guan et al., 1991; Guidato et al., 1996; Sun et al., 1996; Veeranna et al., 1998). Indeed, studies in SOD1 mutant mice have shown that Cdk5 hyperphosphorylates NFs, leading to NF inclusions (Nguyen et al., 2001) and SAPKs phosphorylate NF subunits in response to glutamate treatment, resulting in a slowing of NF transport, in cultured cortical neurons (Ackerley et al., 2000). However, recent studies using liquid chromatography tandem mass spectrometry (LC/MS/MS) have found that the phosphorylation of NF-H from sporadic ALS spinal cord neurons does not differ from control samples (Strong et al., 2001), although it has been proposed that the levels of NF phosphorylation in the cell body compared with the axon may differ and this could affect the assembly and transport of NFs (Miller et al., 2002; Strong et al., 2001).
Role of molecular motor proteins
There is evidence to suggest that defects in the molecular motor proteins which mediate axonal transport may be involved in the slowing of axonal transport in MND. Studies of green fluorescent protein (GFP)-tagged NF-M transport in cultured neurons have shown that transport is dependent on microtubules and the anterograde motor protein kinesin (Koehnle and Brown, 1999; Yabe et al., 1999). Kinesin heavy chain mutations have been found to affect axonal transport in Drosophila (Hurd and Saxton, 1996) and more recently, mice lacking the neuronal-specific kinesin heavy chain KIF5A showed a defect in NF transport, with NF subunits accumulating in neuronal cell bodies (Xia et al., 2003). Furthermore, the transport of tubulin is impaired and axonal levels of kinesin are reduced in SOD1G85R mice months before disease onset (Warita et al., 1999; Zhang et al., 1997), levels of cDNA encoding the kinesin-like protein KIF3B are reduced in the motor neuronal cell line ‘NSC34’ stably-transfected with mutant SOD1 (Kirby et al., 2002) and mutations in KIF1Bβ have been found to cause Charcot Marie Tooth disease type 2A (CMT2A) (Zhao et al., 2001a). Kinesin dysfunction has also been implicated in HSP (Kinesin dysfunction in HSP).
Additionally, components of the retrograde motor protein complex (dynein/dynactin) have been found to associate with NFs and catalyse their transport in vitro (Shah et al., 2000) and mutations in dynactin (p150 subunit) have been found to cause an autosomal dominant form of lower motor neuron disease in humans (Puls et al., 2003). As dynein knockout mutations are lethal in both mouse and Drosophila models, a transgenic mouse model with a targeted disruption of the dynein/dynactin complex was recently engineered to investigate dynein involvement in MND. This was achieved by overexpression of the dynamitin subunit of dynactin (which disassembles dynactin) in postnatal motor neurons, resulting in the development of late-onset progressive motor neuron degeneration and muscle atrophy, thus confirming the critical role of axonal transport in the pathogenesis of MND (LaMonte et al., 2002). Furthermore, mouse models have been utilised to demonstrate that motor neurons are uniquely sensitive to disruption of dynein function and retrograde transport. Mice heterozygous for either of two [ENU]-generated mutations, Loa and Cra1, demonstrate a motor neuron degenerative phenotype similar to the dynamitin transgenic mouse, and positional cloning has revealed that both mutations are in the cytoplasmic dynein heavy chain 1 gene (Dnchc1). Neither mutation seems to affect the localisation of dynein, or its expression levels, but its function is subtly inhibited, affecting motor neurons alone. Retrograde transport of a fluorescently-labelled fragment of Tetanus toxin was found to be significantly reduced in these mice, whereas other functions of dynein such as nuclear motility during cell division, and formation and positioning of the Golgi apparatus, were normal (Hafezparast et al., 2003b; He et al., 2005; Kieran et al., 2005).
Axonal transport defects in hereditary spastic paraplegia (HSP)
One autosomal recessive form of HSP is caused by loss-of-function mutations in the SPG7 gene, which encodes the protein Paraplegin (a member of the ‘ATPases associated with a variety of cellular activities’ (AAA) family). Paraplegin knockout mice display an MND-like pathology and their neurons have large axonal swellings containing organelles and NFs, with impaired retrograde axonal transport (Ferreirinha et al., 2004). The mice also have abnormal mitochondria in spinal cord neurons, and this mitochondrial phenotype correlates with disease onset and neuronal degeneration, which has led to the proposal that mitochondrial dysfunction may underlie defective axonal transport. However, impairment of axonal transport in these mice is only seen after the onset of disease, which indicates that it may not be the primary cause of neurodegeneration (Ferreirinha et al., 2004). Mutations in the neuronal-specific kinesin gene KIF5A have recently been found to cause a form of HSP (SPG10) in humans (Reid et al., 2002), which further implicates defective axonal transport in MND. SPG4/Spastin (an AAA protein, but in a different subgroup to paraplegin) has been found to bind to microtubules, and has been proposed to act as a microtubule-severing enzyme (Charvin et al., 2003; Errico et al., 2002) which suggests that this form of HSP may involve cytoskeletal disassembly/disruption of transport. Interestingly, Spastin abnormally co-localises with kinesin when overexpressed in HEK cells and neurons (McDermott et al., 2003) and recent studies using RNA interference (RNAi) in Drosophila have revealed a role for Spastin in microtubule assembly in synaptic terminals (Trotta et al., 2004). A further microtubule-interacting protein, the novel GTPase Atlastin, is also mutated in juvenile-onset HSP (SPG3A) (Dalpozzo et al., 2003; Zhao et al., 2001b), although there is currently no evidence to suggest that this interaction has any effect on axonal transport. Atlastin shares homology with the Dynamin family of large GTPases, which are involved in molecular trafficking events such as synaptic vesicle recycling and mitochondrial dispersion (Jones et al., 1998; Nicoziani et al., 2000; Smirnova et al., 1998).
A role for glutamatergic excitotoxicity (a mechanism whereby prolonged or excessive exposure to extracellular glutamate leads to cell death) in MND was first implicated by observations of elevated glutamate levels in the cerebrospinal fluid (CSF) of SALS patients (Plaitakis and Caroscio, 1987; Rothstein et al., 1990; Shaw et al., 1995a). The level of neuronal excitation by glutamate is regulated by a number of mechanisms. Two of relevance to MND are NMDA receptor and AMPA/kainate receptor inactivation and Na+/K+-coupled glutamate reuptake by astrocytic transporter proteins (excitatory amino acid transporters; EAAT1-5, also known as glutamate transporters; GLT1-5). There is extensive evidence to suggest an involvement of both mechansims in the pathogenesis of ALS, which is interesting as it highlights the importance of both motor neurons and astrocytes in the disease. The precise molecular mechanisms that lead to excitotoxicity-mediated cell death are not known, although several pathways have been identified that contribute, including disruption of intracellular calcium homeostasis and production of free radicals (For a recent review see (Heath and Shaw, 2002)). Additionally, the motor neurons that are preferentially affected in MND are likely to be particularly sensitive to excitotoxic insults for several reasons, including a high expression of AMPA receptors lacking the GluR2 subunit (this makes them highly calcium-permeable) and a low level of calcium-binding proteins such as calbindin D-28k and parvalbumin, compared with the neuronal populations that are frequently spared in MND such as the oculomotor, trochlear, abducens nerve and Onuf’s nucleus motor neurons (Alexianu et al., 1994; Elliott and Snider, 1995; Ince et al., 1993; Reiner et al., 1995). Indeed, overexpression of parvalbumin in a transgenic mouse model of ALS delayed the disease onset (Beers et al., 2001). Together, these findings support a role for excitotoxicity and disrupted calcium homeostasis in MND.
Role of glutamate transporters
Astrocytes are the principle regulators of extracellular glutamate levels (Rothstein et al., 1996), and deficiency in glutamate uptake by astrocytes appears to play a crucial role in the pathogenesis of ALS. The exact cause of glutamate transport deficiency in ALS is not known, although it has been postulated that the reduction in transport function is caused by a loss of the astrocytic glutamate transporter EAAT2 (also known as GLT1) (Rothstein et al., 1992). In 60-70% of SALS cases, and in SOD1 mutant mouse models, a large reduction in EAAT2 expression (which is not purely due to cell death) is seen at the end-stage of disease (Bristol and Rothstein, 1996; Bruijn et al., 1997b; Rothstein et al., 1995). EAAT2 reduction has also been observed at early stages of disease (before the onset of hind-limb paralysis) in a mutant SOD1G93A rat model and this loss is specific to areas of the spinal cord that contain motor neuron cell bodies (Howland et al., 2002). Furthermore, EAAT2 knock-down, obtained by administration of antisense oligonucleotides in vitro and in vivo, resulted in progressive hind-limb paralysis and motor neuron degeneration in rats (Rothstein et al., 1996). It has been suggested that abnormal mRNA splicing of EAAT2 could underlie reduced expression and function of the protein in ALS (Lin et al., 1998), although several other groups have failed to replicate these findings (Aoki et al., 1998; Flowers et al., 2001). However, no motor neuron loss is seen in EAAT2 knockout mice (although they develop hippocampal pathology and seizures, and most die as juveniles) (Tanaka et al., 1997). This suggests that an overall loss of EAAT2 function does not lead to MND pathology. Additionally, no reduction in EAAT2 expression has been observed in the brain or spinal cord of mice overexpressing SOD1G93A (Deitch et al., 2002). Several studies suggest that the reduction in EAAT2 function may be caused by biochemical changes in transporters already present rather than differences in transporter levels. For example, studies in Xenopus oocytes have shown that the oxidative activity of mutant SOD1 leads to a reduction in EAAT2 function, and this effect can be blocked by the antioxidant Mn(III)TBAP (Trotti et al., 1999). This suggests that EAAT2 may be susceptible to oxidative damage, resulting in decreased glutamate uptake at the synaptic cleft, and excitotoxic damage to motor neurons. Additionally, a mutation in the EAAT2 gene which results in impaired glutamate clearance capacity has also been reported in a single SALS case (Trotti et al., 2001).
Role of glutamate receptors
AMPA receptors are composed of different combinations of four subunits GluR1-4 (or GluRA-D); the GluR2 (GluRB) subunit is of particular importance in determining the calcium permeability of the assembled receptor. An mRNA-editing defect of GluR2 in spinal cord motor neurons from 5 individual SALS cases has recently been reported, which confers increased calcium-permeability on AMPA receptors, leading to cell death (Kawahara et al., 2004). Therefore, the presence/function of the GluR2 subunit and the calcium-permeability of AMPA receptors are believed to play a significant role in excitotoxic pathways in MND. The effect of a reduction in the calcium-permeability of AMPA receptor in motor neurons has recently been investigated in a mutant SOD1 mouse model. This was achieved by crossing SOD1G93A overexpressing mice with a mouse line overexpressing GluR2 specifically in cholinergic neurons (resulting in a large reduction in the calcium-permeability of motor neuronal AMPA receptors). These mice displayed a delay of disease onset with a correlating delay in the formation of abnormal intracellular SOD1 aggregates, as compared with mice overexpressing SOD1G93A alone (Tateno et al., 2004), suggesting that the calcium-permeability of AMPA receptors specifically in motor neurons affects the formation of SOD1 aggregates, leading to MND, in this model. Additionally, the amount of carbonylated proteins (a marker of oxidative stress) in the spinal cord was also reduced/delayed in these mice, and it has been proposed therefore that the protein aggregates seen in the disease may be a consequence of ROS production (Tateno et al., 2004). Complimentary to these findings was a recent study in which a mouse line was created that overexpressed a functionally-modified GluR2 subunit, in which the subunit conferred calcium-permeability on its assembled AMPA receptor but the conductivity of the receptor was unaffected. These mice displayed a phenotype of progressive MND that closely resembles human SALS, and crossing with SOD1G93A transgenic mice resulted in an acceleration of disease progression and decrease in survival, which further supports a role for an increase in calcium permeability caused by defective GluR2 mRNA editing in MND (Kuner et al., 2005).
Involvement of oxidative stress in excitotoxicity
Glutamate transporters are susceptible to oxidative damage, and oxidative modifications of transporter peptides have been reported both in ALS cases and in SOD1 mutant mice (Liu et al., 2002; Pedersen et al., 1998), indicating that the reduction of transporter function in ALS may be caused by oxidative mechanisms. The presence of signs of mitochondrial degeneration in both human ALS cases and in SOD1 mutant mouse models (in which this precedes the onset of motor defects) provides a link between oxidative stress and excitotoxic mechanisms. It has been proposed that lack of mitochondrial function, coupled with the high energy demands of motor neurons, may result in a lowering of the neuron’s membrane potential, resulting in opening of glutamate receptors and influx of calcium, and less glutamate would then be needed to have an excitotoxic effect on the cell. However, a lack of involvement of oxidative stress in downregulation of EAAT2 has been reported in astrocytes (Tortarolo et al., 2004).
Riluzole and evidence against the excitotoxicity hypothesis In support of the glutamatergic excitotoxicity hypothesis, Riluzole, a drug that inhibits glutamatergic transmission, has been shown to delay disease progression in several forms of ALS (Bensimon et al., 1994; Lacomblez et al., 1996). However, the failure of several other antiglutamatergic agents, for example Gabapentin, in human ALS trials (Miller et al., 1996; Miller et al., 2001) suggests that Riluzole may have a novel mode of action. Additionally, a recent study has demonstrated a lack of glutamatergic involvement in ALS, using an in vivo rat spinal cord microdialysis model, and has postulated that large and long-lasting increases of glutamate in the spinal cord do not produce motor neuron hyperexcitation or degeneration (Corona and Tapia, 2004).
Neuroinflammation and autoimmunity
Neuroinflammation has been implicated in ALS due to several findings, including upregulation of the pro-inflammatory enzyme cyclooxygenase-2 (COX-2) mRNA and protein expression, and increased COX-2 activity (measured by increased prostaglandin E2 levels) in SALS spinal cord (Maihofner et al., 2003; Yasojima et al., 2001) and in mutant SOD1 transgenic mice (Almer et al., 2001). Furthermore, a selective COX-2 inhibitor, Celecoxib, protects spinal cord motor neurons from mutant SOD1-mediated cell death by prolonging survival and protecting against microglial activation, astrogliosis and spinal cord neuron degeneration (Drachman et al., 2002; Pompl et al., 2003).
Inflammatory responses also recruit immune mechanisms, the main effectors of which (astrocytes and microglia) are activated in ALS (Hall et al., 1998; Kamo et al., 1987; Kawamata et al., 1992; Schiffer et al., 1996). Indeed, microglial activation occurs before disease onset in mutant SOD1 transgenic mice (Alexianu et al., 2001) and the anti-inflammatory compound minocycline extends survival in mouse models of ALS (Kriz et al., 2002; Van Den Bosch et al., 2002; Zhu et al., 2002), which may occur by inhibiting microglial activation (Tikka et al., 2001; Tikka et al., 2002). Additionally, infiltration of macrophages, mast cells and T cell lymphocytes have been found in SALS spinal cord (Graves et al., 2004; Hayashi et al., 2001; Kawamata et al., 1992; Lampson et al., 1990), which suggests an involvement of both innate and acquired immune responses in MND. Autoantibodies against various components in the CNS have been reported in ALS, although it is still unclear whether this represents an involvement of autoimmunity in the pathogenesis of MND, or if such antibodies have a beneficial effect in the disease. Examples of antibodies found in SALS sera include neuronal, NF subunit and calcium channel antibodies (Couratier et al., 1998; Kimura et al., 1994; Niebroj-Dobosz et al., 1999; Smith et al., 1992). Monoclonal immunoglobulin G (IgG) is also detectable in SALS spinal cord, motor cortex and sera (Duarte et al., 1991; Engelhardt and Appel, 1990). However, anti-inflammatory and immunosuppressive therapies have had little beneficial effect in ALS to date (Drachman et al., 1994).
There is evidence to suggest that neuronal death in SOD1-mediated FALS occurs via apoptotic signalling mechanisms, which is implicated by the presence of DNA fragmentation, decreased expression of the anti-apoptotic protein Bcl-2, and increased expression of the pro-apoptotic protein Bax, in the spinal cords of ALS patients and of transgenic mutant SOD1 mice (Fujita et al., 2002; Martin, 1999; Vukosavic et al., 1999; Yoshiyama et al., 1994). Additionally, expression of Bcl-2 in SOD1G93A mutant mice delays disease onset and increases survival (Kostic et al., 1997) and mutant SOD1 binds to Bcl-2 in mitochondria, which suggests that mutant SOD1 may sequester Bcl-2 into abnormal aggregates and thereby prevent its ability to participate in anti-apoptotic pathways (Pasinelli et al., 2004). Caspase inhibitors have also been effective in delaying onset and prolonging life (Li et al., 2000). Indeed, sequential activation of caspase-1 and caspase-3 (which belong to the family of apoptotic effector cysteine proteases) (Villa et al., 1997) in mutant SOD1 mouse spinal cord motor neurons and astrocytes coincides with onset of motor neuron loss (Li et al., 2000; Pasinelli et al., 2000; Vukosavic et al., 1999).
Three major apoptotic pathways have been the most widely-studied; the mitochondrial (intrinsic) pathway, the death receptor (extrinsic) pathway and the endoplasmic reticulum (ER) pathway. It has been shown that Cytochrome c translocation from mitochondria to the cytosol, and Bax translocation from the cytosol to mitochondria, which are crucial steps in the mitochondrial-dependent apoptotic pathway (Kroemer and Reed, 2000), occurs in the spinal cord of SOD1G93A transgenic mice in parallel with neurodegeneration (Guegan and Sola, 2000). Additionally in these mice, activation of caspase-9 followed by caspase-7, and cleavage of the X chromosome-linked inhibitor of apoptosis protein (XIAP) was reported, further supporting a role of the mitochondrial apoptotic pathway. Involvement of the death receptor pathway has also been identified in MND. Nitric oxide signalling via the death receptor Fas leads to activation of caspase-dependent specific death of embryonic spinal cord motor neurons in culture (Raoul et al., 2002) and another death receptor p75NTR has also been implicated in cell death in mutant SOD1 mice (Kust et al., 2003; Turner et al., 2003).
Although it remains unknown how apoptosis may be triggered in MND, several hypotheses exist. Reactive astrocytes have been shown to promote apoptosis of motor neurons in culture via a nitric oxide and peroxynitrite-dependent mechanism (Cassina et al., 2002) and via nerve growth factor (NGF) in p75NTR-expressing neurons (Pehar et al., 2004). It is also widely believed that mitochondrial dysfunction plays a major role, particularly as mitochondrial abnormality, vacuolation and swelling is observed in ALS patients and FALS mouse models (Dal Canto and Gurney, 1995; Kong and Xu, 1998; Wong et al., 1995), and mutant SOD1 localises to spinal cord and brain mitochondria (Liu et al., 2004; Vijayvergiya et al., 2005), where it impairs Cytochrome c association with the inner mitochondrial membrane and leads to apoptosis (Kirkinezos et al., 2005).
Involvement of cell signalling pathways
Abnormalities in the activity and/or expression of several kinases have been reported in ALS post-mortem tissue, including protein kinase C (PKC), phosphatidylinositol 3-kinase (PI(3)-K), Cdk5 and SAPK (Bajaj et al., 1998; Krieger et al., 1996; Lanius et al., 1995; Migheli et al., 1997; Nagao et al., 1998; Wagey et al., 1998). Furthermore a recent study, using a quantitative proteomics screening technique on human thoracic spinal cord samples from ALS patients and controls, has identified increased expression of several kinases in ALS (Hu et al., 2003b), including PKC, ERK2 and its putative downstream target ribosomal S6 kinase 1 (RSK1), phosphorylated/activated p38 MAPK (p38), protein kinase B (PKB; also known as Akt), SAPK and Cdk5.
Mutant SOD1 transgenic mice have been useful in studying Cdk5-mediated cell death in MND. In SOD1G37R mice, mislocalisation and hyperactivation of Cdk5 has been observed, which correlates with increased production of a truncation product (p25) of its activator p35 (Nguyen et al., 2001). The ways in which Cdk5/p25 activity may cause neurodegeneration are unknown, although aberrant phosphorylation of substrates such as NFs and tau, leading to cytoskeletal alterations/defects in axonal transport, has been suggested as a potential event leading to cell death. Additionally, Cdk5 is involved in several cellular processes, including cell adhesion and synaptic signalling (Bibb et al., 2001; Kwon et al., 2000), alterations in both of which could contribute to neurodegeneration. Interestingly, mice overexpressing Cdk5 and p35 do not show a profound disruption of the cytoskeleton (Van den Haute et al., 2001), unlike mice overexpressing p25 (Ahlijanian et al., 2000), which suggests that p25 is the toxic mediator of events leading to neurodegeneration. Oxidative stress has also been found to promote the generation of p25 from p35 through activation of the Ca2+-dependent protease calpain, resulting in increased Cdk5 activity, increased phosphorylation of NF subunits, and inhibition of axonal transport (Lee et al., 2000; Shea et al., 2004). However, transgenic mice overexpressing SOD1G93A in a p35-null background are not phenotypically different from mice overexpressing SOD1G93A alone (Takahashi and Kulkarni, 2004), which suggests that the production of p25 from p35 and the activation of Cdk5 by p35 is not involved in the SOD1G93A-mediated disease seen in these mice. Therefore, the precise role of Cdk5/p25 activity in ALS is currently unclear.
The activity of p38 is increased in SOD1G93A mouse spinal cord (neurons, astrocytes and microglia), and this occurs before disease onset and correlates with disease progression, although no alterations in its activity have been observed in human ALS tissue compared with controls (Hu et al., 2003a; Hu et al., 2003b; Tortarolo et al., 2003). p38 has been implicated in exitotoxicity, as inhibitors of p38 have been found to rescue cells from glutamate-induced cell death (Kawasaki et al., 1997), and it is also involved in phosphorylation of cytoskeletal proteins and modulating the expression of cytokines, nitric oxide and COX-2 (Ackerley et al., 2004; Guan et al., 1998; Mielke and Herdegen, 2001; Ono and Han, 2000). Although active p38 plays a role in certain apoptotic pathways (Kummer et al., 1997) the p38 activation seen in the SOD1G93A mouse is not thought to lead to apoptosis, due to the lack of ultrastructural apoptotic features in nuclei, DNA fragmentation and activated caspase-3 immunostaining in these mice (Bendotti et al., 2001; Migheli et al., 1999; Tortarolo et al., 2003).
Neuronal cell adhesion molecule L1 (L1CAM) is mutated in an X-linked form of HSP (SPG1) and plays an important role in cell recognition and signalling (Jouet et al., 1994). The L1CAM protein is a member of the immunoglobulin superfamily of cell adhesion molecules and is found primarily in the nervous system where it is involved in cellular processes such as neurite/growth cone guidance and neuronal migration during development, and cell survival (Castellani et al., 2000; Chen et al., 1999; Dahme et al., 1997; Fransen et al., 1998).
Amyotrophic lateral sclerosis 2 (ALS2)/alsin was first identified in 2001 by two independent research groups as a protein that is mutated in rare autosomal recessive juvenile forms of ALS (ALS2) and PLS (jPLS) (Hadano et al., 2001; Yang et al., 2001), and has subsequently been found to cause infantile-onset HSP (IAHSP) (Devon et al., 2003; Eymard-Pierre et al., 2002; Gros-Louis et al., 2003b). Although the exact function of ALS2 is currently unknown, it shares homology with several GTPase-regulating proteins (guanine nucleotide exchange factors; GEFs). It is predicted, therefore, to be involved in various critical cellular processes such as signal transduction, regulation of the cytoskeleton and intracellular trafficking.
Click to read this topic
12/7/2005 3:35:53 PM - TYBURN-spbcjds
Click to read this topic
6/6/2012 2:43:17 PM - 18.104.22.168